CSCI 403 Database Management. Types of Constraints. Implicit Constraints. Application-Based Constraints. Explicit Constraints 9/9/2018

Transcription

1 CSCI 403 Database Management 8 Constraints, keys, indexes Restrictions on tables CONSTRAINTS 2 Types of Constraints Implicit (model-based) Explicit (schema-based) Application-based Implicit Constraints Implied by model of world represented by DB E.g., a column named instructor should contain person names Not (usually) enforced 3 4 Application-Based Constraints Not defined in DB Enforced by applications using DB AKA business rules Often too complex to reasonably implement on database side Example: no employee can have a hire date that is not on the first of some month Explicit Constraints Domain constraints Values in a column must match the type domain Primary key constraints (next section) Foreign key constraints (next section) 5 6 1

2 Keys (Theory) Keys have a very specific definition in relational database theory. KEYS We ll revisit this topic when we get to the theory portion of the class. For now, we ll discuss the practical applications in an SQL database. 7 8 Primary Key Each table can have zero or one primary key Primary key applies to a subset of the columns i.e., one or more identified columns form the key Can be all columns! Does three things: Constrains data in rows to be unique for the combination of columns in the primary key (example next page) Constraints data in primary key columns to be not null Creates an index on the combination of columns Primary Key Example Consider a person table: ssn first last birthdate Wood Carpenter 4/17/ Opal Miner 12/1/ Jane Doe 6/2/1995 SSN would make a good primary key. What about (first, last)? Or (first, last, birthdate)? 9 10 Uniqueness Constraint If we make (first, last) a primary key: cannot have more than one Jane Doe. SSN is probably okay, unless someone has a fake identity If SSN is primary key, we cannot insert a duplicate SSN into the table! SQL Example CREATE TABLE person ( ssn text PRIMARY KEY, birthdate date); INSERT INTO person VALUES (' ', 'Wood', 'Carpenter', '4/17/1975'); OK INSERT INTO person VALUES (' ', 'Evil', 'Impostor', 6/22/1963'); ERROR!

3 Indexes Short explanation: Indexes speed searching by indexed columns E.g., SELECT * FROM person WHERE ssn = ' '; Speeds up most if all indexed columns are searched Can get some speed up by searching first listed column, first two, etc. Longer explanation: Without index, must do linear search look at every row With index, search through an optimized data structure called a Btree look at only a fraction of rows We ll study Btrees later in the semester For the curious: Like a binary search tree but with >2 children per node Performance is log b (n), where b is typically large (e.g. b=100) Creating Primary Keys CREATE TABLE person ( ssn text PRIMARY KEY, birthdate date); \d person Table "public.person" ssn text not null first text last text birthdate date "person_pkey" PRIMARY KEY, btree (ssn) Multiple Column Primary Key CREATE TABLE person2 ( birthdate date, PRIMARY KEY (first, last)); \d person2 Table "public.person2" first text not null last text not null birthdate date "person2_pkey" PRIMARY KEY, btree (first, last) Foreign Key A foreign key defines a constraint between two tables: An FK applies to a subset of columns of one table The FK references a subset of columns of second table FK constraint: Values in FK column(s) must either: Exist in columns in referenced table OR Be NULL Example address street city state zip 123 Mockingbird Ln Golden CO Elm St Arvada CO Main St Durham NC Example in SQL CREATE TABLE state ( state text, code char(2) UNIQUE); -- or often PRIMARY KEY CREATE TABLE address ( street text, city text, state char(2) REFERENCES state (code), zip numeric(5)); state state Colorado North Carolina Virginia code CO NC VA \d address Table "public.address" street text city text state character(2) zip numeric(5,0) Foreign-key constraints: "address_state_fkey" FOREIGN KEY (state) REFERENCES state(code)

4 FK Constraint Example VALUES ('129 West 81st Street', 'New York', 'NY', 10024) OK VALUES ('221B Baker St.', London, NULL, NULL) OK VALUES ('4222 Clinton Way', 'Los Angeles', 'XX', 90204) ERROR Another FK Example VALUES ('129 West 81st Street', 'New York', 'NY', 10024) OK DELETE FROM state WHERE code = 'NY'; ERROR (If we had no addresses in NY, would be fine.) Multiple Column Foreign Key CREATE TABLE foo ( x integer, y date, PRIMARY KEY (x, y)); -- or UNIQUE zz text, xx integer, yy date, FOREIGN KEY (xx, yy) REFERENCES foo(x, y)); MISCELLANEOUS CONSTRAINTS Miscellaneous Constraints NOT NULL disallows NULL values in column UNIQUE imposes uniqueness on a column or set of columns (and creates index) CHECK constraints requires meeting a Boolean condition Example: CREATE TABLE foo ( id integer CHECK (id > 0), name text NOT NULL, x integer UNIQUE ); \d foo Table "public.foo" id integer name text not null x integer "foo_x_key" UNIQUE CONSTRAINT, btree (x) Check constraints: "foo_id_check" CHECK (id > 0) Multiple Column Uniqueness a integer, b text, UNIQUE (b, a)); \d bar Table "public.bar" a integer b text "bar_b_a_key" UNIQUE CONSTRAINT, btree (b, a)

5 Multiple Column Uniqueness a integer, b text, UNIQUE (b, a)); INSERT INTO bar VALUES (1, 'apple'); OK INSERT INTO bar VALUES (2, 'pear'); OK INSERT INTO bar VALUES (1, 'banana'); OK INSERT INTO bar VALUES (3, 'pear'); OK INSERT INTO bar VALUES (1, 'apple'); ERROR Notes Constraints can be combined, e.g. CREATE TABLE foo (x integer UNIQUE NOT NULL ); Foreign key constraints also known as referential integrity constraints Foreign key relationships often mirror a common/likely choice of join Up Next Next lecture: Miscellaneous DDL: default column values, sequences, ALTER TABLE, views, indexes, DROP 27 5